- name
SKILL.md
스킬 파일은 게시자의 저장소에 그대로 있습니다.
복사한 명령을 터미널에 붙여넣습니다. 이 SKILL.md가 해당 에이전트에 설치됩니다.
npx skills add SwotAtmk/infinite-creation --skill h3-prompt-writing -a cursor -y복사한 명령을 터미널에 붙여넣습니다. 이 SKILL.md가 해당 에이전트에 설치됩니다.
npx skills add SwotAtmk/infinite-creation --skill h3-prompt-writing -a claude-code -y복사한 명령을 터미널에 붙여넣습니다. 이 SKILL.md가 해당 에이전트에 설치됩니다.
npx skills add SwotAtmk/infinite-creation --skill h3-prompt-writing -a codex -y복사한 명령을 터미널에 붙여넣습니다. 이 SKILL.md가 해당 에이전트에 설치됩니다.
npx skills add SwotAtmk/infinite-creation --skill h3-prompt-writing -a grok -y복사한 메시지를 Grok Bot 채팅에 보냅니다. Bot이 GitHub에서 이 SKILL.md를 가져와 모든 Bot이 공유하는 스킬에 추가합니다.
Install this skill: https://github.com/SwotAtmk/infinite-creation/tree/HEAD/skills/h3-prompt-writing게시자 제공
view_asset and reuse it across shots — do NOT re-view the same image per shot; use view_shot_references(shot_id) only when a shot introduces an asset you have not yet seen.references/base-en.txt and follow its final prompt structure.references/ref-en.txt and follow its six-section rewrite format.Use integrated_multimodal_description, overall_soundscape, and non_diegetic_music in the order shown in references/base-en.txt.
Ref2VA rewrites use subject_definitions, summary, retention_analysis, detailed_description, overall_soundscape, and non_diegetic_music in that order. Reference labels stay consistent across all sections. In subject_definitions each subject is defined by name + role only — its appearance and clothing are carried by the reference image and must not be spelled out (see references/ref-en.txt §2.1).
Read references/ref-en.txt for label rules, retention analysis, and complete examples.
__MINIMAX_H3_REF_1__ is Ye Xueji's spirit body — and never add age, facial features, face shape, hairstyle or hair colour, skin, build, costume, shoes, accessories, trims, materials or colours. If the task input itself supplies a character's look (asset description, character bible, continuity notes, 「角色一致性」 line), do not copy it into the prompt: state consistency as keep the look of __MINIMAX_H3_REF_N__ / stay faithful to the reference image instead of restating it. Scenes, props and environments are not covered by this rule — describe them normally.slams the table, paces frantically, clenches his fists, leans in intensely, snatches the letter, stumbles back breathless) — and keep the action consistent with that line's emotion. Static characters plus a monotonous read come across as lazy; give each beat visible, purposeful movement.<d>…</d>, after the speaker ID (Sx), explicitly state how the line is spoken — emotion (furious / ecstatic / frantic / explosive / urgent / breathless / anxious / annoyed / amused / sad / nervous / hesitant / calm …), pace (fast / hurried / breathless / rapid-fire / measured / slow / drawn-out / hesitant …), and tone or volume (shout / cry out / whisper / murmur / rising / trembling / flat …). Pick the high-tension end of these lists unless the beat genuinely calls for calm. Never write a bare S1 says, <d>…</d> with no delivery description. Vary delivery and intensity between lines, speakers, and emotional turns within the same shot instead of one flat monotone. Follow the delivery wording shown in references/ref-en.txt (e.g. exclaims with light annoyance, , ) — and prefer its higher-energy phrasings for conflict, urgency, or strong emotion.When H3 cannot generate the whole film in one pass, generation is split into short segments chained by the previous segment's last frame. Drive this from a data contract rather than hand-copied prompts, so segments stay consistent and none of them silently repeats another.
Runner note (this repo): the in-app Agent has no shell tool — it cannot execute
python scripts/.... Writestoryboard.jsonwith the domain tools (set_storyboard/update_shot); thescripts/*.pycompiler helpers are for a human/offline terminal, not a runtime step. Do not attempt to run them.
storyboard.json (see references/storyboard.schema.json). Keep every prompt-bearing value in English.
python scripts/gen_storyboard_llm.py --novel-file novel.md --out storyboard.json drafts the contract with a local/remote OpenAI-compatible LLM and feeds compiler errors back to it until the gates pass — see references/storyboard-from-novel.md. The generator's few-shot style example is passed with --example-file <path> (the repo-shipped sample was moved out of the tree, so point it at a local copy, e.g. --example-file ../../docs/test/storyboard.example.json); it strongly affects output depth and reliability.references/storyboard.schema.json and the prompt.txt migration mapping in references/storyboard-contract.md.docs/test/, not committed): storyboard.example.json (3 shots, with reference images → Ref2VA opener) and storyboard.example-novel.json (a verified 8-shot run over a local test novel → T2VA opener, zero errors/warnings).references/storyboard-contract.md for the field roles, the segment rules, and the contract→H3 section mapping.python scripts/compile_storyboard.py --storyboard-file storyboard.json --dry-run (inspect) or --out segments/ (write one .prompt.txt per segment + segments.json + continuity-ledger.json). When assets live in a library and refs leave paths empty, build the index once with python scripts/build_manifest.py --storyboard storyboard.json --out manifest.json and compile with --manifest manifest.json (see references/manifest.example.json).error before generating: zero-pronoun (no "previous shot / 同上 / 保持原样"), duration coverage, field completeness, executable camera moves, no duplicated segment action, and the fidelity lock (beats[].source: original/added/tbd; --strict-fidelity turns an unconfirmed tbd into an error).Notes:
"allow_repeat": true.duration or split the action into separate shots instead.references/comfyui-node-adapter.md, or a Hub tool) stays outside this skill. The stage-confirmation gate (--confirm) and the fallback ladder (H3 failure → shrink the segment → ask the user) are an interface contract every runner must follow — see references/fallback-policy.md.<Picture 1>, <Video 1>, <Audio 1>) across every section.Write MiniMax H3 video generation prompts for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. Use when rewriting multimodal requests into H3 prompt structures, composing integrated_multimodal_description, overal
---name: h3-prompt-writingdescription: Write MiniMax H3 video generation prompts for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. Use when rewriting…---# skills/h3-prompt-writing/SKILL.md개요
Write MiniMax H3 video generation prompts for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. Use when rewriting multimodal requests into H3 prompt structures, composing integrated_multimodal_description, overall_soundscape, and non_diegetic_music, aligning keyframes, or defining reference labels for images, videos, and audio. Also covers storyboard-contract driven multi-segment generation: drafting a storyboard.json from a novel/script with a local or remote LLM, then compiling it into consistent per-segment H3 prompts (Ref2VA/T2VA first segment + FL2VA continuations) with no repeated segments and no positional or identity drift.
FAQ
게시자에게서 skills/h3-prompt-writing/SKILL.md를 다운로드하거나 경로를 복사해 저장소를 엽니다. PlugMCP는 스킬을 목록에 올릴 뿐 파일을 호스팅하지 않습니다.
게시자가 https://github.com/SwotAtmk/infinite-creation에서 관리합니다.
Claude, Cursor, Codex, Grok Build는 게시자의 저장소에서 추가한 SKILL.md를 읽을 수 있습니다. Grok Bot은 채팅에서 스킬의 GitHub 폴더 링크를 보내면 설치합니다.
says in a casual young male voice with a playful tone and an easy conversational pacereplies … with an amused cadencerefs